Flash memory, attributing to its convenience, high memory density and ideal reliability, has been a research hotspot in non-violate storage devices. Currently, development of flash memory is promoted by motivation of chasing higher memory density. However, due to structure limitation, memory densities of conventional flash memories may no longer be increased by simply shrinking the device dimensions.
Generally, flash memory devices can be classified into two types: gate-split flash memories and stacked flash memories. Compared with stacked flash memories, gate-split flash memories, due to their specific structures, may have better performance in programming and erasing. For example, gate-split flash memories may have higher programming efficiency, and word line structures thereof can prevent “over erase”. Therefore, gate-split flash memories are used broadly. However, compared with stacked flash memories, gate-split flash memories have one additional word line, which may increase the chip area. To increase memory density, normally there is a need to improve structures of gate-split flash memories.
Some existing flash memory units may have two silicon nitride layers functioning as float gates in each bit unit. Therefore, in each flash memory unit, there may be stored 4 bits of data, which may greatly increase the memory density. However, manufacturing processes of these flash memory units may be complicated.